No label defined (Q3706429)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 3934325
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 3934325 |
Statements
1985
0 references
stochastic approximation
0 references
adaptive control
0 references
convergence
0 references
regression
0 references
0.8290007710456848
0 references
The stochastic approximation method is very often used for adaptive control and optimization problems where the measurement errors probably are far from the classical independent situations. In order to cover the possibly correlated noise and to weaken the conditions imposed on the regression function a so called ordinary differential equation method was developed by \textit{L. Ljung} [IEEE Trans. Autom. Control AC-22, 4, 551-575 (1977; Zbl 0362.93031)] and \textit{H. Kushner} and \textit{D. Clark} [Stochastic approximation methods for constrained and unconstrained systems (1978; Zbl 0381.60004)] and it works quite successfully, but in this method it is assumed that the estimates given by the algorithm are a priori bounded. Obviously, this is an undesirable restriction. To remove it some efforts have been made recently by the author and \textit{Y. M. Zhu} [Stochastic approximation with randomly varying truncations (to appear)] where by use of a randomly varying truncation technique the boundedness assumption is no longer required and the convergence to the value sought for holds as before. In this paper we present first a convergence result which is slightly stronger than that given in the last mentioned paper. Then we apply it to the optimization problem and we can see that the conditions needed for convergence of the algorithm to the extremum of a regression function probably are the weakest.
0 references